isolated by filtering, washing Kith water and then acetone, and drying, was shown to be 2-amino4hydroxypteridine6-carboxylic acid (111) by comparison of the ultraviolet absorption curve with that of an authentic sample.8 9,lO-Dimethylpteroic Acid (IV) .-This was prepared similarly to Y above, except th'tt p-methylaminobenzoic acidJ was uied (see Table I for ultraviolet absorption and b:ological data). A n a l . Calcd. for C16HltN603.0.5H20: C, 55.1; €I, 1- 01; N, 24.1. Found: C, 54.7; H, 4.99; N, 24.0. N-[ { N-[ 1-(2,4-Diamino-6-pteridyl) -ethyl] -N-methylamino)-benzoyl] -glutamic Acid, ("4-Amino-9,lO-dimethylpteroylglutamic Acid") (VI).-This coinpound wa5 prepared by the method of IValler, et d . , I using 2,4,5,6tetraminopyrimidine sulfate,Q 2,2,3-trichlorobutyraldehyde and p-methylaminobenzoylglutamic acid.a A mixture of 31.6 g. of the crude, 9 g. of lime, and 2 1. of water was heated a t 60" for forty minutes. After ?ltration, the cake was washed with 600 ml. of water a t 60 . The cornbined filtrate and wash liquors were adjusted to pH 10.811.0 with 10% aqueous zinc chloride. The mixture was filtered, and the filtrate T ~ adjusted S to p H 3 5-4.0 with hydrochloric acid. After cooling to 5", the precipitate was filtered, washed with water and slurried in 1500 ml. of Rater containing sodium hydroxide to give PI3 11-11.5. After heating it a t 60" for ten minutes, the PH of the sol:tion was reduced to 7, while cooling the mixture to 20 , and filtered. The filtrate n a s adjusted to pH 3.5-4.0 with hydrochloric acid and cooled t o 5'. After filtration, the cake was slurried in 1 1. of water containing enough lime to obtain PH 8.5-9 a t 60" and filtered hot with 1.5 g. of char-
coal. The filtrate was adjusted to p: 3.5-4.0 with hydrochloric acid and cooled to 0-5 overnight. The yellow-orange microcrystalline material which separated was isolated by filtration and was analytically pure. The biological and ultraviolet absorption data are shown in Table I. A n a l . Calcd. for CzlHzaNsOb.2HzO: C, 50.0; H, 5.55; K, 22.2. Found: C,50.4; H, 5.42; N,21.5.
Acknowledgment.-We are indebted to hlr. Richard I,. Shepard for technical assistance in the preparation of certain of these substances, to AIiss Kuth Xbbott for the ultraviolet absorption data, and to X r . 0. Sundberg and coworkers for the microanalyses. Summary N- [4- ( [1-(2-Amino-4-hydroxy-6-pteridyl)ethyl]-amino ] -benzoyl] -glutamic acid (11), designated herein as 9-methylpteroylglutamic acid, has been syntheszed and isolated in pure form. Corresponding analogs which have also been prepared are 9,lO-dimethylpteroylglutamicacid (V), 4-amino-9,lO-dimethylpteroylglutamic acid (VI), and 9,lO-dimethylpteroic acid (IV). These compounds exhibit antagonism for pteroylglutamic acid in the growth of S.faecalis R. ROUNDBROOK,NEW JERSEY RECEIVED SEPTEMBER 27, 1948
@) Traubc, Ber., 37, 4.545 (1901)
[CONTRIBUTION FROM
THE
DEPARTMENT O F CHEMISTRY, UNIVERSITY O F WISCONSIN]
Mechanical Properties of Substances of High Molecular Weight. V. of Polyisobutylene Solutions in Various Solvents'
Rigidities
BY J. N. ASH WORTH^ AND JOHN D. FERRY
Many concentrated solutions of linear polymers behave as highly viscous liquids in steady flow, but possess solid-like rigidity in small oscillating deformations ; transverse waves can be propagated in them a t audio frequencies. From the wave length of these waves the rigidity can be calculated, and from the damping the mechanical loss, or the dynamic viscosity, can be derived. Measurements of this sort on solutions of polystyrene in xylene have been reported previouslymJ In seeking t o understand the mechanism of rigidity in polymer solutions, i t is desirable to study a variety of molecular types. I n this paper, measurements are reported for a non-polar polymer for which complications due to intermolecular forces should be at a minimum : polyisobutylene. As in the earlier work on polystyrene, the dependence of rigidity on requency, concentration, and temperature has been studied. ,41so, four different solvents have been employed. (1) Supported 111 part by t h e Research Committee of t h e Graduate Srhool of the University of U'isconsin from funds supplied b y t h e \\ isconsin Alumni Research Foundation, and in part by a grant from Research Corporation ( 2 ) Present address Kohm and Haas C o , Philadeli>hi L I,+. 'I) T I) liLrr\, Trrrq Tor R Y 11 64 1 . { 3 (1942)
Materials The polyisobutylene was kindly furnished by Dr. John Rehner, Jr., of the Esso Laboratories. It was designated 303-92-1. From measurements of viscosities of dilute solutions in &-isobutylene, the intrinsic viscosity in this solvent at 20' was found to be 2.85 (g./lOO cc.)-I, corresponding to a viscosity-average molecular weight of 1,200,000 on the basis of the interpolation equation of F l ~ r y . ~ Intrinsic viscosities in the solvents used for the rigidity studies were also determined a t 20' and 30' ; these are given in Table IV below. The details of the intrinsic viscosity measurements are presented elsewhere. In most cases, the polymer was used with&t purification. However, i t contained a small amount of tiny fibrous foreign particles which, though negligible iq quantity, were optically anisotropic and interfered somewhat with optical measurements of waves. Accordingly, a portion of the polymer was dissolved in toluene a t a concentration of about five per cent., clarified by filtration through Whatman No. 4 filter paper on a (4) P J Flory, ibid., 66, 372 (1943). ( 3 ) J N.Ashworth, P h D. Thesis, TJnirersity of Wisconsin, 1948
Feb., 1949
RIGIDITIESO F
POLYISOBUTYLENE SOLUTIONS I N
Buchner funnel, completely precipitated by methyl alcohol, and dried in vacuo a t room temperature for twelve hours and a t 60’ for four hours. This preparation, hereafter referred to as clarified polymer, afforded the best measurements of damping. The n-heptane was “99 mole per cent.” sold by the Phillips Petroleum Corp. Xylene was obtained from Mallinckrodt and Merck, and chlorobenzene and isooctane (2,2,4-trimethylpentane) from the Eastman Kodak Company. The solvents were used without further purification. Preparation of concentrated solutions was expedited by rotating bottles containing polymer and solvent a t about 12 revolutions per hour. Concentrations were determined by dry weight after heating samples in vacuo for .twenty-four hours a t 110’. Concentrations were expressed both as weight per cent. and as g./cc. The latter values were calculated assuming additivity of the volumes of polymer and solvent, taking the density of polymer as 0.91 and’those of the solvents from the ‘‘International Critical Tables,” and assuming that the thermal expansion coefficients of solutions were the same as those of the respective solvent^.^
Method The apparatus and procedure were the same as previously d e ~ c r i b e d with , ~ , ~ some minor modifications. The Strobotron was mounted outside the air thermostat, and its illumination of the cell was improved by a reflector and a condensing lens. Photographic recording of the wave pattern was substituted for the former method of visual observation with horizontal and vertical micrometer adjustments. The rather faint pattern was satisfactorily photographed on Eastman Kodak Co. CTC plates or Contrast Process Panchromatic cut film, the latter affording a contrast between light and dark Babinet bands which was an improvement over visual observation. The photographs of wave patterns were projected by an enlarger on cross-section paper, and the contours of the wavy Babinet bands were traced in pencil. An example is shown in Fig. 1. Similar projections of photographs of the undeviated Babinet bands, when superimposed, permitted measurements of the displacements from which damping data were calculated. The wave length was taken as the average of ten or more measured distances between successive maxima or minima on each photograph. There was no perceptible change in wave length with distance from the vibration source. As in the previous studies on polystyrene, the damping was within experimental error exponential; a plot of In 6 (6 being the double refraction in the solution observed a t the maxima and minima of the stroboscopically immobilized wave) against x , the distance from the vibrator plate, was a (6) J. D. Ferry, Rru. Sci. Instvumenfs, 12, 79 (1941).
VARIOUS SOLVENTS
623
rl
U Fig. 1.-Tracing of wave pattern in 15.7% polyisobutylene in chlorobenzene, 17.1O , 400 cyclesjsec. The vibrator plate is indicated at left.
straight line.’ The reciprocal slope of the line is critical damping distance. Values of 6 can be calculated from the equation6 tan 6 = tan A sin 2 cy, where A is the relative retardation corresponding the Babinet line displacement, and CY is the angle between the axes of the Babinet and the polarizer. However, a simpler procedure was followed; since sin 2 a is small (in our case cy = 2’14’), tan 6 is nearly equal to 6, and tan A is proportional to 6 ; it follows that the reciprocal slope of a plot of In tan A against x is XO, and the critical damping distance is obtained without calculating 6 directly. The wave rigidity,’ G, was calculated as X V p , where X is the measured wave length, v the frequency, and p the density. The latter value was calculated assuming additivity of the volumes of polymer and solvent. The dimensionless damping index,gXjxo, was used to characterize the severity of the damping. Different sizes of cells and vibrator plates were used in four combinations (Fig. 2). Comparisons of the same solutions with different cells and plates XO,the
a
C
b d Fig. 2.-Cell-plate combinations (top view). Cell sizes: a, 1.0 X 3.5 cm.: b, 4.0 X 4.0 cm.; c and d, 2.5 X 6.0 cm. Plate widths: a, b, and c, 0.8 cm.; d, 2.5 cm. (7) T h e first maximum or minimum often deviated materially. This is t o he expected from theoretical calculations8 of wave propagation in which t h e finite dimensions of t h e cell are taken into account, T h e calculations show t h a t t h e value of xo determined as outliped here is within a few per cent., and hence well within experimental error, of t h e value which would characterize t h e damping of a plane wave in a medium of infinite extent.9 (8) F. T. Adler, W. M. Sawyer a n d J. D. Ferry, unpublished work. (9) J. D. Ferry, W. M. Sawyer and J. N. Ashworth, J . Polymer Sci., 2, 593 (1947).
J. N. ASHWORTH AND
624
revealed no differences in G and X/XO exceeding experimental scatter. Most measurements were made with combinations (c) and (d), which provided room for longer wave trains. The widest frequency range was achieved with combination (d). The cell used in (c) and (d) was of aluminum with glass windows cemented with litharge and glycerol. I n some cases the dynamic rigidity G' and the dynamic viscosity TJ', which are useful for comparison with results of other experimental methods and for theoretical interpretation, were calculated by the equation9
where w is 27r times the frequency. Results Dispersion of Rigidity.-In experiments where a frequency range of fiye or sixfold was achieved, a small increase in G with increasing frequency was observed. Representative data for clarified polymer in chlorobenzene solutions are given in Table I (concentrations expr-ssed in weight per cent). The dependence of G on the frequency can be approximated by straight lines of equal slope (Fig. 3).
JOHN
D. FERRY
Vol. 71
G IN
TABLE I CHLOROBENZENE SOLUTIONS AT 25"
DISPERSION OF Frequency, cycles/sec.
100 110 125 160 200 210 250 320 400 500 630 800
6, dyn/sq.
cm., X 10-4 Concn. = 11.2% 13.3%
8.6%
..
1.82
0.85
..
.88
1.80
.89
1.80
.96 .99 1.00 1.12
1.83 1.93 1.89 1.93 2.24
..
..
I
+
-
..
TABLE I1
G OF %-HEPTANE SOLUTIONS Concn. Frequency polymer, range % cycles/s'ec.
T;mp.,
E , dyn./ C,
C.
g./cc.
In other experiments, where the frequency range was limited t o three or fourfold, the dispersion of was within experimental scatter (5 to 10%). It was, in any case, so small that the dispersion function appears to be essentially a plateau in the range from 100 to 1000 cycles/sec. for our solutions, Accordingly, to determine the dependence of G on concentration and temperature, values measured a t different frequencies were simply averaged. Dependence of Rigidity on Concentration and Temperature.-Average values of G, together with the frequency range covered in each
sq. cm., x 10-4
E/ca
x
10-1
13,'l
125-320
-6.9 1.3 10.5 18.5 28.4 35.2
0.0955 .0946 .0935 .0926 .0916 .0908
0.66 .62 .59 .57 .57 .51 Av.
15.4
160-400
1.0 10.9 18.8 27.5 35.4
0.112
1.12 8.0 1.05 7.7 1.12 8.4 1.06 8.2 1.00 8.0 Av. 8.1
-7.8 0.9 10.9 19.3 27.8 34.6
0.149
-5.0 1.3 11.1 19.5 28.4 36.5
0.188 .187 .I85
-6.0 1.4 10.9 18.9 28.3
0.226 ,224 ,221 .219 .217
11.2
200 400 600 800 Frequency, cycles/sec. Fig. 3.-G plotted against frequency for chlorobenzene solutions at 25". Figures opposite the lines denote concentrations in weight per cent.
..
4.31 5.11 4.90 5.39 5.34 5.41
case, are given for solutions in n-heptane a t different concentrations and temperatures in Table 11. At constant weight concentration, the rigidity decreases somewhat with increasing temperature,
13.3
8.6
.. 4.67 4.38
..
15.7
t
..
2.90 3.00 3.15 3.16 3.13 3.36
.. ..
..
..
2.73 2.54
..
..
15.7%
.. ..
20.1
25.0
125-500
125-630
c
29.7
160-800
.111 ,110 ,109 .lo8
.148 .146 .145 .143 .I44
.183 .181 ,179
7.5 7.3 7.2 7.2 7.4 6.8 7.2
2.88 8.7 2.82 8.7 2.59 8.4 2.98 9.7 2.59 8.7 2.50 8.7 Av. 8.8 5.56 5.48 5.49 5.33 5.09 4.88 Av.
8.4 8.4 8.8 8.8 8.6 8.6 8.6
9.01 7.8 8.98 7.9 8.23 7.6 8.41 8.0 7.91 7.7 Av. 7 . 8 Over-all average 8 . 1
Feb., 1949
RIGIDITIESOF
625
POLYISOBUTYLENE SOLUTIONS IN 17ARIOUS SOLVENTS
about 0.3% per degree. I n the case of polystyrene-xylene solutions, where a somewhat more pronounced decrease was ~ b s e r v e dthe , ~ temperature dependence of could be described by the = AGeQGIRT, with QQ = 1500 cal. relation The results of Table I1 can also be fitted by this relation, with QQ = 550 cal. However, for polyisobutylene, the temperature dependence a t constant weight concentration can be entirely attributed to the change in volume concentration due to thermal expansion ; when rigidities are compared a t constant volume concentration, the temperature dependence v_anishes. Proportionality of G to the third power of c, the volume concentration (g./cc.) was anticipated from the behavior of polystyrene.l0 The last column of Table I1 shows that for polyisobutylene in n-heptane G / c 3is constant for each solution over a temperature rangejrom about -7 t o 35'. The average values of G/c3 for each solution are also close together, and no trend with increasing concentration is apparent.12
Data for the other solvents are given in detail in the the.$s5 on which this paper is based. In each case G decreased about 0.3% per degree a t constant weight concentration, but G / c 3 was independent of temperature and ~0ncentration.l~ The values of G / c 3 averaged a t different temperatures for each solution are given in Table 111. In Table IV, the over-all averages of G/c3 for the four solvents are correlated with the respective intrinsic viscosities a t 20 and 30°. The intrinsic viscosities show that the solvents are arranged in the order of decreasing solvent power; the temperature dependence, which is probably significant only for chlorobenzene, also indicates that this is a somewhat poorer solvent than the others. The differences are on the whole slight, but i t is clear that the order of decreasing intrinsic viscosity is the order of increasing rigidity. TABLE IV AVERAGE VALUESOF
TABLE I11
10.3" 8.9 11.5" 9.3 11.5b 9.7 15.3" 9.6 15 3" 9 6 18.5" 8.9 18.7"" 11.1 18.7"" 9.7 9.4 18.7de 11.1 24.5' 8.6 25. lee 9.1 25.3" 9.5 28.7" 7.9 33.5c 7.9 Iqooctane 10. 7df 12.2 15.4" 9.0 10.1 19.7d 10.2 26. 4df 9.1 Chlorobenzene 8 . 6def 11.9 11 .2dEf 10.8 10.9 13.3dej 10.2 15. 7de 10.5 a Cell-plate combination a. * Cell-plate combination b. Cell-plate combination c. Cell-plate combination d. Clarified polymer. f Measured at 25' only. (10) Proportionality t o ca was shown for the rigidity of polystyrene solutions after a calculated low-frequency component of rigidity had been subtracted from t h e measured values.11 However, the low-frequency component was so small t h a t the relation holds also for t h e total rigidity (see Fig. 7). (11) J. D. Ferry, THISJOURNAL, 64, 1330 (1942). (12) Since t h e frequency range represented by the average 2; changes somewhat with Concentration (Table II), this does not quite represent a test of t h e third power relation a t constant frequency. Values interpolated from Fig. 3 a t constant frequency and plotted logarithmically against log c give slopes ranging from 2.7 t o 3.0. However, these small differences are too close t o experimental error t o attempt t o describe a t present.
INTRINSIC VISCOSITIES
(dyn./sq.'cm.) (g./cc.) - 3 X IO-'
Solvent
VALUES OF G/c3 FOR SOLUTIONS IN XYLENE,ISOOCTANE, A N D CHLOROBENZENE Solvent Cpncn., wt. % G/ca x 1 0 4 AV.
Xylene
AND G/c3
[?I, 20' 2.68 2.63 2.45 2.15
8.1 9.4 10.1 10.9
n-Heptane Xylene Isooctane Chlorobenzene
[vi, 30'
2.60 2.67 2.36 2.35
Damping.-In almost all cases, the damping index, X/xo, varied within the rather narrow range from 0.6 to 1.2; Except in a few cases, the experimental scatter did not permit a reliable study of the dependence of damping index on concentration, temperature, or nature of solvent; as a first approximation, it was independent of all these variables. With increasing frequency, X/xo generally decreased. The best measurements, on chlorobenzene solutions, are given in Table V and plotted against the logarithm of the frequency in Fig. 4. Dynamic Rigidity and Viscosity.-From the data in Tables I and V, the dynamic rigidity G' and the dynamic viscosity 7' of the chlorobenzene TABLB V DAMPING INDICESIN CHLOROBENZENE AT 25' Frequency cycles/sec.'
125 160 200 210 250 320 400 500 630 800
11.2%
1.14 0.97
..
1.00 0.79 .77 .64 .73
..
..
Damping index, X/xo Concn. =
13.3%
15.7%
..
..
..
1.20
..
0.96 .96 .72 .69 .66 .70
..
.. .. 0.86 .94 .55 .82 .74 .64
(13) Since t h e thermal expansion coefficients of the solvents are all very nearly one-tenth per cent. per degree, ca decreases threetenths per cent. per degree, accounting exactly for the observed decrease in rigidity.
J. N. ASHWORTHAND JOHN D. FERRY
626
Vol. 71
1.4 71.2
-
0
$ 1.0 x$ 0.8 -
15.7
.*e
.-c 0.6 -
13.3 11.2
F
2 0.4 -
a
L
o.2 0 100 250 630 Frequency (logarithmic). Fig. 4.-Damping indices for chlorobenzene solutions, plotted against the logarithm of the frequency. Figures opposite the curves denote concentrations in weight per cent. The ordinate scale refers to the 15.7% curve; the curves for 13.3% and 11.2% have been shifted downward 0.2 and 0.4 unit, respectively. The curves are calculated for retarded Maxwell elements, using the constants of Table VI.
solutions were calculated by equations (1) and (a), and they are plotted against the logarithm of the frequency in Figs. 5 and 6. The rigidity G' is from three t o ten per cent. smaller than G. The viscosity q' is of the order of a few poises and falls off rapidly with increasing frequency; this is the usual behavior found for concentrated polymer solutions by direct measurements with transducer methods.14 By contrast, the viscosity in steady flow of these solutions is of the order of hundreds to thousands of poises.
ol,
,
,
,
111.2
100 250 630 Frequency (logarithmic) .. Fig. 6.-Dynamic viscosities of chlorobenzene solutions, plotted against the logarithm of the frequency. Figures opposite the curves denote concentrations in weight per cent. The curves are calculated for retarded Maxwell elements, using the constants of Table VI.
where G" values for solutions are compared with G' for the undiluted polymer. In most cases the accuracy of X/XO is insufficient for an adequate calculation of q', although its order of magnitude can always be estimated from Eq. 2, since X/XO almost never exceeds the range 0.6-1.2. 7.0 I
I
I
,/j ,
I /'
6.0 . /
16 M
3 5.0 15.7 4.0
13.3 I
t--"
11.2
t .
100 250 630 Frequency (logarithmic). Fig. 5.-Dynarnic rigidities of chlorobenzene solutionsr plotted against the logarithm of the frequency. Figures opposite the curves denote concentrations in weight per cent. The curves are calculated for retarded Maxwell elements, using the constants of Table VI.
Since G' never differs from by more than 10% in our experiments, the two rigidities can be taken as equivalent for some purposes, as in Fig. 7, (14) T.L. Smith, J. D.Ferry and F. W. Schremp, J . A p p l . Physics, in press.
I
I
-0.8 -0.4 0 Log c. Fig. '?.-Log plotted against log c. Solid lines (drawn with slope of 3), in descending order: polyisobutylene in chlorobenzene, isooctane, xylene, and heptane ; polystyrene in xylene (points from ref. 3). Rectangle: G' of pure polyisobutylene (ref. 16). -1.2
e
Discussion Representation of Frequency Dependence by a Mechanical Model.-A simple model which can imitate approximately the frequency dependence of mechanical properties of concentrated polymer solutions in the audio frequency range is a retarded Maxwell element,gwith a viscous dashpot (rs) in series with a parallel combination of spring (G) and dashpot ( 7 ~ ) . The three model constants are frequency-independent. In the 10, when q p
